How to use satellite imagery to visualise changes in landscapes

And how those changes can help you chronolocate an event.

(Click here to watch and listen to the video version of this blog entry)

It is absolutely phenomenal knowing that we live in a time where anyone with an internet connection can easily gain access, for free, to satellite imagery gathered from satellites orbiting Earth. Every day millions of images are collected, capturing every nook and cranny of our planet. This invaluable data helps professionals (and hobbyists) to better understand our planet and make data driven decisions.

In OSINT, access to such data can help geolocate an image or video and, in some cases, narrow it down to a time period. This ability to lock down an event to a specific time frame is colloquially referred to as chronolocation.

Imagine that you are looking at a photo of something happening in a small village in the middle of nowhere. The media claims the event happened on X day but satellite imagery can tell you otherwise. By analysing what is happening on the background of the image or video and comparing it with satellite images of the area, you may be able to narrow down the event to a very small time frame. This sometimes can be enough to fight disinformation and propaganda.

I have been playing around with Sentinel Hub EO Browser for a while now. This website compiles a “complete archive of Sentinel-1, Sentinel-2, Sentinel-3, Sentinel-5P, ESA’s archive of Landsat 5, 7 and 8, global coverage of Landsat 8, Envisat Meris, MODIS, Proba-V and GIBS products in one place.”
It has an extensive amount of data available, for free, and it is very easy to use.

So how does it work?

First thing we need to do is decide what we want to investigate. I chose to focus on wildlife fires as it is something that is easily visible from satellite images and also fairly straightforward to compare with imagery from earlier dates.
So then I had to find out where on our planet we currently have wildfires. For that I used the website RiskMap with only the “Fire” option selected.
I selected the earliest date on the time frame (shown below) in order to find fires that have been active for a while. I also chose the latest date to be of a few days ago so I could make sure there was enough time for the satellites to gather and put images available on the Sentinel Hub website.

I was drawn to Argentina where the area of Corrientes seems to have been facing wildfires for a few weeks.

Moving around the area using the Sentinel Hub EO Browser, I managed to find a location that was likely quite affected by the wildfires. You can also get there by inputing the following coordinates at the top right of the screen: -27.7692, -57.3216.

I have decided to stick to the Sentinel-2 satellite images as it is also used for monitoring burnt areas which is exactly what we are trying to do.

Once we click “Search” we have to select which data set we want to use. I tend to just pick one with very small cloud coverage to allow better visibility. The Sentinel-2 L2A only has a 6.9% cloud coverage and its last image was gathered yesterday, on the 22nd of February 2022, so it is perfect.

Now that we can see the satellite image of the location, the wildfire destruction does seem substantial. We can even spot a few spots still aflame, circled in red below.

In order to check out a different time we navigate to “Date” and select one from the list of available snapshots. It is quite useful to also limit the amount of cloud coverage to make sure that the area we want to inspect is actually visible. We can see for example a satellite image from the 8th of January 2022 below.

We can even play a bit more with timelapses and create a little animation. For that you will need to register for a (free) user account. Then simply select the “create timelapse animation” icon on the right and centre the area you want to view. Press the blue button in the middle to start creating it.

I have selected to view one photograph per day, limiting it to images with a maximum cloud coverage of 25%, with the start point on the 29th December 2021 and the most recent one from the 22nd February 2022. We can then select or deselect any photo that is (or not) useful for our animation.

I have picked 7 different images from that time frame and then simply saved the animation by clicking the “download” button.
You can watch the final result below.
We can then infer that the fire reached the outskirts of the nearby village between the 2nd and the 7th of February 2022.